Urban Dreams of Migrants: A Case Study of Migrant Integration in Shanghai

Unprecedented human mobility has driven the rapid urbanization around the world. In China, the fraction of population dwelling in cities increased from 17.9% to 52.6% between 1978 and 2012. Such large-scale migration poses challenges for policymakers and important questions for researchers. To investigate the process of migrant integration, we employ a one-month complete dataset of telecommunication metadata in Shanghai with 54 million users and 698 million call logs. We find systematic differences between locals and migrants in their mobile communication networks and geographical locations. For instance, migrants have more diverse contacts and move around the city with a larger radius than locals after they settle down. By distinguishing new migrants (who recently moved to Shanghai) from settled migrants (who have been in Shanghai for a while), we demonstrate the integration process of new migrants in their first three weeks. Moreover, we formulate classification problems to predict whether a person is a migrant. Our classifier is able to achieve an F1-score of 0.82 when distinguishing settled migrants from locals, but it remains challenging to identify new migrants because of class imbalance. This classification setup holds promise for identifying new migrants who will successfully integrate into locals (new migrants that misclassified as locals).

💡 Research Summary

This paper, “Urban Dreams of Migrants: A Case Study of Migrant Integration in Shanghai,” presents a large-scale quantitative analysis of migrant integration in urban China using a one-month dataset of telecommunication metadata from Shanghai. The dataset, provided by China Telecom, encompasses approximately 54 million users and 698 million call logs, augmented with user demographics (age, gender, birthplace) and the geographic locations of call towers.

The core of the methodology involves categorizing users into three groups based on birthplace and calling activity: Locals (born in Shanghai), Settled Migrants (born outside Shanghai with call logs in the first week of the study period), and New Migrants (born outside Shanghai with no calls in the first week but calls in subsequent weeks). This yields a final sample of 1.7M locals, 1.0M settled migrants, and 22K new migrants. The analysis focuses on features extracted from weekly mobile communication networks (e.g., demographics of contacts, network structure, call behavior) and geographical mobility patterns (e.g., activity radius, movement distance).

The study’s key findings are multi-faceted. First, in a static comparison, systematic differences emerge among the groups. Locals’ communication networks are highly localized, with about 70% of their contacts being other locals. In contrast, migrants rely more on fellow townsmen (people from the same home province), with this fraction being around 30% for new migrants and surprisingly increasing to about 35% for settled migrants. Settled migrants also exhibit the largest social networks (highest degree) and the widest geographical activity radius, wider than both locals and new migrants. Spatial distribution maps reveal a pattern reminiscent of “white flight,” with locals relatively concentrated in the city’s periphery and migrants in the center.

Second, a dynamic analysis over the three-week observation period for new migrants shows a process of integration. The features of new migrants, such as their townsmen contact ratio and activity radius, gradually become more similar to those of settled migrants over time, while the characteristics of locals and settled migrants remain stable. However, the rate of this convergence slows in the final week, suggesting that not all new migrants integrate at the same pace and some may face difficulties.

Third, the researchers formulate classification tasks using the extracted features. A classifier trained to distinguish settled migrants from locals achieves a high F1-score of 0.82, demonstrating that the behavioral patterns of these two groups are distinctly different. Classifying new migrants proves more challenging due to class imbalance. The paper suggests that this classification framework holds promise for identifying new migrants who are likely to integrate successfully—those who are “misclassified” as locals by the model. This could serve as an early indicator for policymakers to target support.

In conclusion, this work offers a novel, data-driven lens into the complex social process of urban migrant integration. By leveraging digital behavioral traces, it quantifies differences in social networks and spatial mobility, reveals the incremental nature of integration, and provides a methodological foundation for predictive analytics that could inform more nuanced urban and migration policies.